HYBRID QUANTUM SYSTEMS: RESONATOR STATE PREPARATION WITH RYDBERG ATOM BEAMS
Robert Jirschik
Supervisor: Prof. Stephen Hogan
Co-supervisor: Prof. Sougato Bose
Department of Physics and Astronomy
University College London
20.05.2019
A dissertation submitted to University College London for the degree of Doctor of Philosophy Declaration
I, Robert Jirschik, confirm that the work presented in this thesis is my own. Where information has been derived from other sources, I confirm that this has been indicated in the thesis.
Date: ...... Signature: ......
2 Abstract
In this thesis theoretical studies of hybrid quantum systems composed of solid-state superconducting microwave resonators, mechanical oscil- lators, and gas-phase Rydberg atoms are described. The thesis begins with an overview of the current state of the field of hybrid quantum in- formation processing. An introduction to the physics of Rydberg atoms and the preparation processes of circular Rydberg states follows, includ- ing a review of the associated literature. Three main new research results are then presented. These include (1) numerical studies of static electric dipole interactions in strongly polarized gases of Rydberg atoms. The understanding and characterization of these interactions is essential to maximize the quantum-state preparation procedures required for the hy- brid systems studied in the thesis; (2) analytical and numerical studies of quantum-state preparation and cooling in coplanar superconducting mi- crowave resonators using beams of atoms in circular Rydberg states; and (3) studies of coupled Rydberg-atom—mechanical-oscillator systems. The results presented in each of these areas are of interest for hybrid quan- tum information processing and quantum computation, and optical-to- microwave photon conversion for quantum communication.
3 Impact Statement
The research reported in this thesis is of interest in several research areas. Approaches to the preparation of quantum states of resonator systems in a controlled manner as presented here have applications in several quantum technologies. For example, the selective preparation of the vacuum state of the resonator can be used to increase the fidelity of quantum state transfer protocols and thus be applied in the context of quantum communication. The study of electrostatic dipole-dipole interactions within ensembles of Rydberg atoms described in this thesis further the understanding of many- body physics in strongly polarized gases. The combination of this study and the resonator state preparation results could lead to new quantum measurement techniques to characterize atomic ensembles, as well as the realization of quantum non-demolition measurements of gas-phase atoms and molecules, including exotic species such as antihydrogen and positro- nium. In the long term, the results presented here constitute a stepping stone in the development of a hybrid quantum information processing architecture with transmission-line microwave resonators controlled and manipulated using Rydberg atoms. This could lead to the development of new, robust quantum simulators for studying complex condensed matter problems that cannot be solved using classical computation technology.
4 Acknowledgements
There are too many people to thank for all the support I received during my time as a PhD student. I shall give it a try here. First and foremost I would like to thank my supervisor, Prof. Stephen Hogan, for giving me the opportunity to work with him. His passion for science, his dedication, interest and enthusiasm with which he approaches his work all have been a great motivation for me. I could not have finished this thesis without his support, ideas, suggestions and discussions. I would also like to thank my Co-supervisor, Prof. Sougato Bose, for his cre- ative input and help. Another thank you to my Diplomarbeit-supervisor Prof. Michael Hartmann. My overwhelmingly positive experience during my research year changed my perspective on physics, and encouraged me to pursue a PhD. A special thanks to my parents, Kathrin and Dirk, who have been encour- aging me to follow my passions without a worry. They are amazing role models. I cannot thank them enough for their advice and moral support. I would also like to thank a few PhD students and Postdocs who have passed through our office over the years, or with whom I had the plea- sure of discussing our work: Hoda Hossein-Nejad, Patrick Lancuba, Ed- ward O’Reilly, David Holdaway, Richard Stones, Valentina Zhelyazkova,
5 Valentina Notararigo, Stefan Siwiak-Jaszek, Adam Deller, James Palmer and Alex Morgan. My apologies to everyone else I have interacted with at UCL and did not mention here, your contributions are all appreciated. And finally, a big thank you to all my close friends. It seems unfair to mention only some of them by name, but a special thank you to my clos- est friend Felix for his unrivaled reliability, and the two people with the biggest influence on me during my time in London, Renaud and Olivier. Without their distractions this endeavour would not have been even half as fun.
6 Contents
1 Introduction 22
2 Rydberg Atoms 29 2.1 Hydrogen atoms in Rydberg States ...... 29 2.2 Helium atoms in Rydberg states ...... 33 2.3 Atoms in Circular Rydberg States ...... 33 2.3.1 Helium Atoms in Circular Rydberg States ...... 40
3 Electrostatic Dipole Interactions in Polarized Rydberg Gases 44 3.1 Experiment ...... 47 3.2 Results ...... 51 3.3 Simulations ...... 53 3.4 Conclusions ...... 58
4 Superconducting Transmission Line Resonator 60 4.1 Quantization of a 1D Transmission Line Resonator ...... 61 4.2 Dissipation from a 1D Transmission Line Resonator . . . . . 65
5 Cooling a transmission-line resonator with a beam of Rydberg atoms 72
7 5.1 The atomic beam as an engineered quantum reservoir ...... 73 5.2 Resonator De-Excitation ...... 81 5.2.1 Steady State Solution ...... 85 5.2.2 Time-Resolved Solution ...... 94
6 Atomic Beam Interaction: Numerical Approach 100 6.1 Modeling the Beam of Atoms ...... 101 6.2 Stochastic Dynamics ...... 106 6.2.1 The stochastic process as a series of time steps . . . . 107 6.2.2 The dynamics for a constant number of atoms . . . . 108 6.2.3 A beam of atoms interacting with the resonator . . . 109 6.2.4 Ensemble average of the trajectories ...... 111
7 A Microwave Resonator-Nanomechanical Oscillator–System driven by an Atomic Beam 117 7.1 Derivation of the Hamiltonian ...... 118 7.2 Microwave Resonator-Mechanical Oscillator–Subsystem . . 122 7.3 Microwave Resonator-Mechanical Oscillator-Atom–System . 125 7.4 The Atomic Beam as an engineered environment ...... 130 7.4.1 Steady State of the system interacting with an atomic beam ...... 135 7.4.2 Considering a Thermal Environment ...... 138
8 Conclusions 143
A Probability of N Atoms interacting with the Resonator 164
8 List of Figures
1.1 Overview of hybrid quantum systems. Each individual sys- tem is positioned according to its excitation frequency and coherence time, with possible couplings between them in- dicated by arrows. From [65]...... 24
2.1 The helium atom in a high n Rydberg state. The outer elec- tron "sees" the ion core and inner electron as one central field with charge Z = +1. The Coloumb potential associated with the He2+ nucleus is screened by the inner electron...... 34
2.2 The effect of the quantum defects δn,` on the binding energy of the atomic states for non-hydrogenic atoms compared to the hydrogen atom. From [53]...... 35 2.3 Adiabatically switching off the electric field, F~, leads to a rotation of the vector operator ~λ. This is equivalent to the evolution of the upper state’s parabolic ellipse, with angular
momentum m = 0, to a circular trajectory with m = Lz = n 1. 39 − 2.4 The energy level diagrams for steps 2-4 in the preparation process of circular Rydberg states for n = 70 in Helium as described in the text. From [80]...... 41
9 3.1 (a) Schematic diagram of the photoexcitation and detec- tion region of the experimental apparatus. The position of laser photoexcitation and the approximate spatial extent of the Rydberg atom ensemble in the y dimension during microwave interrogation are indicated by the red circle and dashed rectangle, respectively. (b) Sequence of events in each experimental cycle, including laser photoexcitation, Rydberg state polarization, microwave interrogation, and state-selective electric field ionization (not to scale). From [128]...... 48
3.2 Stark structure of the triplet m` = 0 Rydberg states of He with values of n ranging from 70 to 72. Experiments were performed following laser photoexcitation of the 70s state in zero electric field and the subsequent polarization of the
1 excited ensemble of atoms in fields of (i) 18 mV cm− , (ii)
1 1 358 mV cm− , and (iii) 714 mV cm− as indicated. From [128]. 50
10 3.3 Experimentally recorded (continuous curves) and calculated (dashed curves) spectra of transitions between the Stark
3 3 states that evolve adiabatically to the 1s70s S1 and 1s72s S1 levels in zero electric field. The experiments were performed at low (blue curves) and high (red curves) Rydberg atom number densities, and spectra were recorded in polarizing
1 1 electric fields of (i) 18 mV cm− , (ii) 358 mV cm− , and (iii)
1 714 mV cm− , for which µ70 = 380 D, 7250 D, and 12250 D
(µ72 = 470 D, 8650 D, and 14050 D), respectively. The cal- culated spectra are presented (a) without, and (b) with ef- fects of the local polarization of the medium on the macro- scopic dielectric properties accounted for. The relative mi- crowave frequencies on the horizontal axes are displayed with respect to the transition frequencies [(i) 37.2413 GHz, (ii) 37.1068 GHz, and (iii) 36.8094 GHz] recorded at low Ry- dberg atom number density. From [128]...... 52 3.4 Experimentally recorded (continuous curves) and calculated (dashed curves) dependence of the n = 70 n = 72 spectra → of the most strongly polarized ensemble of atoms, for which
µ70 = 12 250 D and µ72 = 14050 D, on the Rydberg atom density. From [128]...... 57
11 4.1 Simple LC-circuit electrical oscillator: (a) In order to derive a Lagrangian where the single degree of freedom is the charge, q, the usual sign convention for the flow of the current, I, from q to q can be used. (b) The sign convention in the − text, where the positive and negative signs of the charge, Q, have been flipped with respect to the current, leads to the magnetic flow as the single degree of freedom in the Lagrangian describing the LC-circuit. From [41]...... 61 4.2 The resonator-waveguide system and its lumped circuit ana- logue. The resonator and waveguide can be considered as a sum of harmonic LC-circuits, where coupling capactiances at each end impose boundary conditions...... 66 4.3 Comparison of the here calculated Q-factor and the mea- surement by Goeppl et al. Both curves show qualitatively similar behavior, whereas the precise values differ because the details of the experimental parameters of the waveguide could only be estimated from the published information. . . 70
12 5.1 The population transfer rates An, Bn and Cn among three consecutive Fock states in the resonator. It can be seen that
Bn corresponds to a de-excitation of the resonator from the
Fock state n to the Fock state n 1 , whereas Cn describes | i | − i an excitation from the state n to the state n + 1 . The | i | i total rate of de-population of the Fock state n is therefore | i An = Bn+Cn. Together with the competing transfer rates Bn+1
and Cn 1 populating the n Fock state, the rate of change of − | i the population of that state n can be calculated according | i to Eq. (5.28)...... 83 5.2 Distribution of the photon occupation number for a funda- mental frequency of ν = 18.78 GHz. The red dots mark the occupation probability of the Fock state n of the resonator | i in the non-equilibrium steady state. The Planck distribu-
tion for an effective temperature Teff = 360 mK is plotted (blue line), showing the qualitative fit with the steady state photon distribution...... 86
13 5.3 Depletion of the fundamental mode of the resonator. The thinner red bars represent the occupation probability dis- tribution of the Fock states n at T = 4 K. This results in | i an average photon occupation number of n = 3.96. The h ith thicker blue bars represent the photon number distribution after the interaction with an atomic beam with a fixed atom- resonator interaction time τ¯, leading to a thermal photon number distribution with an average number of photons n = 0.122...... 87 h iss 5.4 The dependence of the 0 Fock state fidelity on the detun- | i ing of the atomic transition frequency and the resonator resonance frequency of the selected fundamental mode. . . . 89 5.5 The dependence of the resonator ground state fidelity on the atom-resonator coupling strength. A coherent beam (solid blue line) reaches local minima for specific values of the interaction strength, one being at g = 62.82 MHz. The minima represent trapping states. The incoherent atomic beam (dashed red line) does not exhibit such behavior, with a roughly constant fidelity of 0.85 for values of g 80 F0 ≈ ≥ MHz...... 90 5.6 Photon number distribution corresponding to the annoma- lous trapping state that arrises for an atom-resonator cou- pling strength of g = 62.82 MHz...... 92
14 5.7 The dependence of the steady state average photon num- ber , n , on the resonator quality factor, Q. The product h iss rτ 1 ensures the approximation of at most one atom in- teracting with the resonator is fulfilled...... 93 5.8 Dependence of the average photon number n of the nonequi- h i librium steady state of the resonator on the rate of atoms crossing the resonator with different average interaction times τ¯ as indicated. The continuous curves represent the values for r that fulfill the approximation rτ 1, where the probability of having two or more atoms interact with the resonator at the same is less than 0.25%. The dashed curves represent the region in which this approximation breaks down because the probability that two or more atoms in- teract with the resonator at the same time is not negligible anymore...... 95 5.9 Time-dependent evolution of the average photon number n of the resonator. The initial state of the resonator with a h i fundamental frequency of ν = 18.78 MHz is assumed to be a
thermal state at Tenv = 4 K. The atoms on resonance, ∆ = 0,
6 1 cross the resonator at a rate r = 10 s− and couple to it for a constant interaction time τ¯ = 50 ns with a coupling strength g = 26.66 MHz...... 98 5.10 Time resolved reheating of the resonator...... 99
15 6.1 A schematic diagram of the experimental setup, allowing for more than one atom to interact with the microwave res- onator at any given time t...... 101 6.2 The flow chart of the numerical approach. Within each time step ∆t the number of atoms leaving the microwave res- onator is calculated (orange arrows) after having interacted
over a time tint. This depends on how many atoms remain within the resonator, and how many new atoms enter the microwave resonator (black arrows), based on their arrival
probability Pn(∆t). The black arrows from subsequent or- ange atom number boxes, have been omitted for clarity of the sketch...... 103 6.3 The probability distribution representing the number of atoms interacting with a resonator over a time T = 2 µs, with an average interaction time of τ¯ = 10 ns and a rate of r = 40 atoms/µs. The simulation was carried out for 500 trajectories...... 106
16 6.4 Schematic diagram of the average photon occupation num- ber of the microwave resonator with atoms being determin- istically monitored for a single realization. The moments atoms enter the resonator are marked by grey, dashed grid- lines. Three atoms enter at a time t = 0, the state of one of them being displayed by the dashed blue curve. Two atoms leave and one enters at t = 500. At a time t = 1000, the atom 1 leaves the resonator, and a different atom 2 enters, whose dynamics are displayed by green dashed curves, and two additional atoms are introduced. At t = 1500 the same atom remains and its state is displayed, while 2 new atoms enter, one of which is being displayed by the red dashed curve. . . 110 6.5 Comparison of the analytical and numerical method of treat- ing the coupled atom-resonator system in a parameter regime in which both approaches are valid. The rate of Rydberg
6 1 atoms crossing the resonator is r = 10 s− , the constant in- teraction time is τ = 50 ns, and the interaction strength is g/(2π) = 4.25 MHz...... 113 6.6 The effect of a dense beam of atoms prepared in the lower of two circular Rydberg states on the thermal photon occu- pation number in the resonator. The rate of Rydberg atoms crossing the resonator is r = 15 MHz, the constant inter- action time is τ = 50 ns, and the interaction strength is g/(2π) = 4.25 MHz...... 114
17 6.7 Two calculated trajectories for individual atoms compared to the average of 100 simulated trajectories, displaying the quantum jumps in the photon field induced by an atom entering the resonator and being excited...... 115 6.8 Comparison of the photon probability distribution of the method a) studied by Sarkany et al. [99] and b) the method investigated in this chapter with an atomic rate of r = 145 · 6 1 10 s− and an average interaction time of τ¯ = 10 ns. Both concepts lead to a sharp peak around n = 0 and a signifi- cantly lower average effective photon number in the steady state...... 115 6.9 Time-dependence of the average resonator photon number. An average steady state photon number of n & 0.1 can be h iss reached within a time of t 1 µs, utilizing approximately ≈ 150 Rydberg atoms to effectively cool the resonator...... 116
7.1 (1) The dependence of the eigenenergies of both polariton- modes is displayed when the detuning between the mi-
crowave resonator and the laser drive ∆˜ 0 is decreased from
2 ωm to 0 ωm by varying the laser drive frequency. (2a) and (2b) show the relative amplitudes of the photonic and phononic excitation...... 124
18 7.2 The first 3 eigenenergies of the Hamiltonian Hi depending on the detuning between the microwave resonator detuning
and the mechanical oscillator ∆ = ∆˜ ωm. The detuning 0 − between the microwave resonator and the laser drive ∆˜ 0 is
decreased from 2 ωm to 0 ωm...... 127
7.3 The quasiparticle-composition of the eigenenergies (a) ωx,
(b) ωy and (c) ωz in Equ. (7.29) of the interaction Hamiltonian,
Hi, that are plotted in Fig. 7.2...... 129 7.4 The effective cooling process for the polariton excitations
α†α + β†β . Note that the x-axis is inversed in order to h iss h iss illustrate the effect for an increase in the rate of atoms pre- pared in the lower circular Rydberg state. The inset shows the same calculation with a linear y-axis...... 137 7.5 The dependence of the two polariton branches on the rate
of the atomic beam,r, for Nm = 40. Because γr/γm = 5, the average polariton numbers of the initial, thermal state are
α†α 4.74 (mainly photonic mode) and β†β 9.17 h ith ≈ h ith ≈ (mainly phononic mode)...... 142
A.1 The dependence of the probability of more than five atoms interacting with the resonator at the same time on the prod- uct of r τ. The dashed lines indicate the maximum value of · the product r τ for which the error is 0.5%...... 168 · ≤
19 List of Tables
1.1 Important properties for different systems that can be used as qubits [49, 57, 80, 81, 83, 121]...... 23 1.2 Important properties for different types of resonators [57, 42, 81, 63, 25]...... 25
2.1 Comparison between some of the properties of low- and high-angular-momentum Rydberg states...... 37
4.1 The values used for the theoretical calculation of the Q-factor as published by Goeppl et al. [42]...... 70
20 "He helps you to understand He does everything he can Doctor Robert" - The Beatles
21 Chapter 1
Introduction
In recent years the development of quantum technologies has made con- siderable progress. The ability to control the quantum dynamics in ex- periments with atoms and trapped ions has been greatly enhanced and culminated in the award of the Nobel Prize in Physics to Serge Haroche and David J. Wineland in 2012 for their work in cavity quantum electrody- namics [47, 119]. In addition to this discipline within quantum optics, the development of superconducting quantum electrical circuits [124, 31, 113] in the last two decades and the proposal to, i.e., employ such systems as quantum simulators for light-matter interactions [12], has further been aid- ing the progress of the field of quantum information processing because of their scalability and control. Other systems have also been emerg- ing as promising candidates as quantum bits for quantum information processing due to their long coherence times, such as ensembles of elec- tron spins [120], nitrogen-vacancies (NV) in diamonds [7], and quantum dots in semiconductors [90]. Additionally, the use of micro- and nanome- chanical oscillators, i.e., in the form of microtoroidal resonators, carbon
22 nanotubes or membranes, and the ability to reach the quantized regime of their mechanical motion have been offering new possibilities for tests of fundamental quantum physics [15, 74, 25], and for quantum information processing [96]. All of these individual quantum systems, that contain qubits and resonators, offer distinct advantages and disadvantages, as outlined in Table 1.1 and Table 1.2.
n = 70 Optical Electronic SC Transmon Rydberg atom Ion spin ensembles qubit Energy 20 GHz 105 GHz 1 10 GHz 5 10 GHz ∼ − ∼ − ∼ − gap f 106 GHz Coherence 600 µs 1 35 s 1 ms 1 s 5 20 µs ∼ ∼ − ∼ − ∼ − time T2 Coupling 25 MHz 10 kHz 100 Hz 150 MHz ∼ ∼ ∼ − ∼ rate g/2π 1 MHz
Table 1.1: Important properties for different systems that can be used as qubits [49, 57, 80, 81, 83, 121].
At present there is considerable interest in the development of hybrid approaches to quantum information processing. These involve combining two or more distinct quantum systems, where, for example, one is being employed as the quantum memory, i.e. a quibit, and coupled to another quantum system, i.e. a resonator, to perform gate operations, in order to take advantage of the most favorable aspects of
23 Figure 1.1: Overview of hybrid quantum systems. Each individual sys- tem is positioned according to its excitation frequency and coherence time, with possible couplings between them indicated by arrows. From [65]. each. These may include, for example, coherence times, scalability or the timescales on which states can be prepared or manipulated. Within the last decade, hybrid approaches to quantum information processing have been proposed and realized in various ways. One such proposal is the use of ensembles of cold polar molecules acting as a long-lived quantum memory coupled to a circuit QED setup consisting of a superconducting cooper pair box strongly coupled to a stripline resonator [94]. Another proposal is the magnetic coupling of a Bose-Einstein condensate of rubid- ium atoms to the nanomechanical motion of a cantilever [110]. Examples
24 for the experimental realization of strongly coupled hybrid quantum sys- tems are the coupling of electron spin ensembles to a superconducting transmission-line [100] as well as the emergence of circuit quantum acous- todynamics where a superconducting qubit has been coupled to trapped on-chip acoustic wavepackets [76].
Optical Coplanar wave- Si Nanobeam cavity guide resonator Mechanical oscillator Frequency 105 GHz 2 8 GHz 4 GHz ∼ ∼ − ∼ range Coupling strength 100 MHz > 300 MHz 1 MHz ∼ ∼ g/2π Quality > 106 105 108 105 ∼ − ∼ factor Q
Table 1.2: Important properties for different types of resonators [57, 42, 81, 63, 25].
In this context the work described in this thesis is related to the investiga- tion of chip-based approaches to exploiting Rydberg atoms in hybrid quan- tum information processing [122, 98, 72, 89, 88]. The long-term objetives of this work are to coherently couple gas-phase atoms in highly excited Ry- dberg states to solid-state superconducting qubits. The coupling between the two systems will be mediated, resonantly or near-resonantly through coplanar superconducting microwave resonators. Because of the possibil-
25 ity of implementing fast quantum gates (g/2π 100 MHz) when using ≈ superconducting qubits and the inherent scalability of chip-based struc- tures [58], superconducting microwave circuits appear to be ideal solid- state candidates for the realization of quantum approaches to computation and simulation. However, superconducting qubits suffer from compar- itively short coherence times (τ 10 µs). They are therefore not ideally ≈ suited as long coherence-time quantum memories. With this in mind efforts are being directed toward the development of hybrid approaches to quan- tum information processing in which rapid state manipulation could be performed in these solid-state circuits while a gas-phase sample of atoms, molecules or ions is used as a long coherence-time memory [94]. From this perspective, atoms in highly excited Rydberg states are well suited to act as the gas-phase component of such a hybrid system. Rydberg states can exhibit comparitively long coherence times ( 100 ms) and because of ≈ their large electric dipole moments for transitions at microwave frequen- cies they are compatible with coupling to microwave circuits. A realization of such a hybrid system would open up possibilities for several new appli- cations such as long-coherence-time quantum memories [104], or enabling the coupling between optical and microwave photons [13] in a control- lable medium for microwave-to-optical conversion [38]. Other possiblities include using Rydberg atoms coupled to superconducting circuits for pho- ton number state preparation as has been done in large three-dimensional cavities [111], and long distance quantum entanglement distribution [27] in spatially separated quantum circuits coupled via transmission lines. In approaching the goal of strongly coupling gas-phase Rydberg atoms to
26 superconducting microwave circuits, a chip-based architecture for com- prehensive control of the translational motion and internal quantum states of Rydberg atoms has recently been developed [66, 67, 51]. This archi- tecture allows ensembles of Rydberg atoms to be accelerated, decelerated and transported in travelling electric traps located directly above copla- nar electrode structures that are suitable for microwave transmission. The same electrode structures can act as coplanar microwave resonators and electrostatic traps for Rydberg atoms located above them [68]. In this thesis, theoretical studies are reported upon which those experi- mental developments will build. They include (1) studies of static electric dipole interactions within ensembles of Rydberg atoms which play an im- portant role in the preservation of coherence over long timescales in these samples, (2) analysis of a scheme to use beams of Rydberg atoms to cool selected resonator modes, and (3) the extension of this cooling scheme to cool selected modes of other solid-state systems, such as nanomechanical oscillators. In Chapter 2 the theoretical description of hydrogen atoms in Rydberg states and its extension to helium atoms is reviewed. Then, the proper- ties of Rydberg atoms in circular states are discussed, the general method to prepare atoms in such circular states is reviewed, and extended to the preparation of helium atoms in circular Rydberg states. In Chapter 3 the role of static electric dipole-dipole interactions between helium Ryd- berg atoms, which lead to mean-field energy shifts in an atomic ensemble, is investigated. The general quantization procedure of superconducting transmission line resonators is then discussed in Chapter 4, and the open
27 quantum systems approach to quantifying the dissipation within these ar- chitectures when coupled to a thermal bath such as a coplanar waveguide used for measurements is then outlined. Chapter 5 deals with the general theory of quantum reservoir engineering and how it is applied to a beam of Rydberg atoms interacting with a transmission line resonator. Within this context an original cooling scheme for hybrid quantum systems with Rydberg atoms is presented and an analytic solution is given for the diag- onal elements of the resonator density matrix. Chapter 6 improves upon this idea by introducing a numerical method that provides an even more effective cooling mechanism as well as more insights due to the calcula- tion of the full density matrix of the resonator. Chapter 7 introduces an additional hybrid quantum system including a nanomechanical oscillator, whose quantum states are controlled via quantum reservoir engineering by an atomic beam. Finally, in Chapter 8 open problems and the outlook of this research are discussed.
28 Chapter 2
Rydberg Atoms
Rydberg atoms are atoms excited to states with very high principal quan- tum number n. They have been of importance in the derivation of Bohr’s quantum mechanical orbital model of the electronic structure of atoms, and have played a key role in the development of quantum optics, quan- tum information and many-body physics due to the possibility of realiz- ing strong electric dipole coupling to single microwave photons [18] and strong dipole-dipole interactions [6] between pairs of Rydberg atoms. In this chapter some of the properties of Rydberg atoms, and methods of Rydberg state preparation relevant to the topic of this thesis are described.
2.1 Hydrogen atoms in Rydberg States
The simplest atoms to be theoretically described are the hydrogen and hydrogen-like atoms. These are two-particle quantum systems, which can be approximately described as consisting of a positively charged nucleus and one electron. The electrostatic interaction between these two particles
29 is expressed by the Coloumb potential
Ze2 V(r) = , (2.1) −4π0r where the charge of the nucleus is Ze, e is the charge of the electron, 0 is the electric permittivity of the vacuum, and r is the distance between the electron and the nucleus. The wavefunction of the electron Ψ can be described via the Schrödinger equation [28] ! ~2 Ze2 2 Ψ = EΨ, (2.2) −2µ∇ − 4π0r where ~ is the reduced Planck constant, and µ = meM is the reduced me+M mass of the electron-nucleus system with me and M being the mass of the electron and nucleus, respectively. The energy of the electronic eigenstates is denoted by E. Switching from Cartesian coordinates (x, y, z) to spherical coordinates q r = x2 + y2 + z2, (2.3)
z θ = arccos , (2.4) r y φ = arctan , (2.5) x
Eq.(2.2) can be seperated1 into radial and angular components. Hence, the wavefunction is factorizable as [11]
Ψ(r, θ, φ) = R(r) Y(θ, φ). (2.6) ·
If both the radial and angular parts of the Schrödinger equation are equal but with opposite signs, they cancel each other, leading to a nontrivial
2 1The Laplace operator transforms as 2 = 1 ∂ r2 ∂ + 1 ∂ sin θ ∂ + 1 ∂ ∇ r2 ∂r ∂r r2 sin θ ∂θ ∂θ r2 sin2 θ ∂ϕ2 30 solution. By defining this separation constant [45] as `(` + 1), two self- consistent eigensystems follow:
Angular equation • " ! # 1 ∂ ∂ 1 ∂2 sin θ + Y(θ, φ) = `(` + 1). Y(θ, φ) sin θ ∂θ ∂θ Y(θ, φ) sin2 θ ∂φ2 − (2.7) The solutions to this equation are the fatorizable normalized spherical
harmonic functions Y`,m(θ, φ) = f`,m(θ)gm(φ), where the azimuthal quantum number m is introduced as another separation constant analogous to the factorization of the electron wavefuntion above. The generating function for the spherical harmonics is r m (2` + 1)(` m)! imφ Y`,m(θ, φ) = ( 1) − P`,m(cos θ)e − 4π(` + m)! (2.8)
= Y`,∗ m(θ, φ), −
with m 0, and where P m(cos θ) are the associated Legendre poly- ≥ `, nomials.
Radial equation • " ! 2 2 ! # d 2 d 2µr Ze r + 2 E + `(` + 1) R(r) = 0. (2.9) dr dr ~ 4π0r −
By substituting y(r) = rR(r), one arrives at a form of a well known equation whose orthonormal set of analytical solutions is related to
a the associated Laguerre functions Lb(x). The solution for the eigen- states is [44] s !3 !` ! Zµr 2Zµ (n ` 1)! 2Zµr 2`+1 2Zµr − nmea0 Rn,`(r) = − − 3 e Ln ` 1 nmea0 2n[(n + `)!] nmea0 − − nmea0 (2.10)
31 2 2 where a0 = 4π0~ /(mee ) is the Bohr radius and the bound states are categorized by the principal quantum number n = 1, 2, 3,... and the angular momentum quantum number 0 ` n 1. ≤ ≤ −
The eigenenergies follow from the solution to the radial equation and give the Rydberg formula Rhc 13.6 eV En = = − (2.11) − n2 n2 where the Rydberg constant for an electron bound to an infinitely heavy positive point charge can be expressed in terms of the fundamental con- stants Z2e4m Z2~2 R = e = (2.12) hc 2 2 2 2 16π 0~ a0me Several properties of low-` Rydberg states follow from this treatment of the structure of the hydrogen atom. Especially interesting are the scaling laws of the atomic properties with the principal quantum number n. For higher Rydberg states, i.e., large n, the binding energy of the electron decreases
2 with n− . The energy difference between adjacent eigenstates decreases
3 by n− . Hence, the radiative lifetime of such transitions increases with n3. The large electric dipole transition moments, ~p, for transitions where
2 ∆n = n0 n = 1, scale as p n0`0 e zˆ n` n . This is a consequence of − ± ∝ h | · | i ∝ the large orbital radius of Rydberg states, which scales with n2. All these properties mean that atoms in high-n Rydberg states are well suited to applications in quantum information processing [36].
32 2.2 Helium atoms in Rydberg states
For helium atoms where one electron is excited to a high energy state the Rydberg formula Eq. (2.11) needs to be adjusted by introducing the quantum defect δn`. This empirical parameter accounts for the fact that the electron excited to a Rydberg state interacts with what can be viewed as an ion core whose nuclear charge is screened by the surrounding second electron, as can be seen in Fig. 2.1. Consequently, the Coloumb potential of the ion core is perturbed as it is no longer spherically symmetric. It is also possible that the inner electron is polarized by the outer electron in the Rydberg state. Therefore, the difference in binding energy for the hydrogen case can be accounted for by replacing the principal quantum number n by an effective quantum number [36] n∗ = n δn . The adjusted − ` Rydberg formula for helium is
RHe hc En = 2 (2.13) −(n δn ) − ` R µHe where R = ∞ is the Rydberg constant for the helium atom. As can be He me inferred from Fig. (2.2), the deviation from our model of the ion core and the inner electron as one composite ion core with charge Z = +1 is bigger for the outer electron in Rydberg states closer to the positive core. Thus, the quantum defect, δn`, decreases for Rydberg states with higher quantum numbers n and `.
2.3 Atoms in Circular Rydberg States
Rydberg states which are of particular interest as qubit states in hybrid cavity QED are those which possess maximum angular momentum and
33 Electron in high n, ℓ Rydberg state
Inner electron and ion core
Figure 2.1: The helium atom in a high n Rydberg state. The outer electron "sees" the ion core and inner electron as one central field with charge Z = +1. The Coloumb potential associated with the He2+ nucleus is screened by the inner electron.
magnetic quantum numbers ` and m`, respectively, for a particular value of n. Such states are called circular Rydberg states and exhibit unique properties. The first of these are their long radiative lifetimes. Since the Einstein A coefficients describing the transition rate from an initial state i to a final state f , i.e., the decay rate of an excited state, can be written as a sum of the Einstein A coefficients of the decay into all possible final states. The lifetime of the excited state is then the inverse of this decay rate, and
34 Figure 2.2: The effect of the quantum defects δn,` on the binding energy of the atomic states for non-hydrogenic atoms compared to the hydrogen atom. From [53]. thus 1 2ω3 − 1 X i f `max ~ 2 τ = P = 3 p f i , (2.14) A f i 30~c 2`i + 1| | f f where A f i is proportional to the cube of the energy (frequency) difference between the initial and final states, and the square of the transition dipole moment [33]. In the case of circular Rydberg states, the excited state has ` = m = n 1. ` − Due to the electric dipole selection rules allowing up to first order only transitions satisfying `0 = ` 1 and m0 = m or m0 = m 1, it follows that − ` ` ` ` ±
35 decay can only occur to final states `0 = n 2. Consequently, the initial − excited circular state can only decay to one unique final state, n0 = n 1, − `0 = n 2, m0 = m 1. Therefore, circular Rydberg states are a very − | `| | `| − good approximation to 2-level systems. Within the dipole approximation this quasi two-level characterization only breaks down on long timescales since the lower n0 = n 1 circular state can itself decay to the next higher- − lying circular state with n00 = n 2. However, in an electromagnetic − environment that is appropriately engineered such that no vacuum mode exists at the frequency of the n 1 n 2 circular-state to circular- − → − state transition, this additional decay channel can be eliminated [59]. The transition energy between the two states making up the quasi two-level
3 system is En,n 1 n− , while their electron dipole transition moment scales − ∝ as n 1, n 2 ~r n, n 1 n2. Therefore, it follows for the radiative lifetime h − − | | − i ∼ of circular Rydberg states
1 5 τ f i 3 3 2 2 = n . (2.15) ∝ (n− ) (n )
Hence, circular Rydberg states exhibit greatly enhanced radiative lifetimes compared to low-` states with the same values of n. The properties dis- cussed above are listed in Tab. (2.1) for two examples of Rydberg states of Helium, showing an increased fluorescence lifetime in the case of circular Rydberg states when compared to states with low orbital angular momen- tum.
Circular Rydberg states cannot be directly photoexcited from an atomic ground state because of the selection rules for electric dipole transitions. However, two methods have been developed and implemented experi-
36 Table 2.1: Comparison between some of the properties of low- and high- angular-momentum Rydberg states. mentally to prepare them. The first of these involves adiabatic microwave transfer from an initially laser photoexcited low-` (low-m`) Rydberg state with the absorption of approximately n microwave photons [60, 69]. The second method to produce circular Rydberg states is known as the crossed-fields method, and was first proposed by Delande and Gay [29]. By applying a constant magnetic field crossed perpendicularly with an ad- justable electric field, one is able to very efficiently prepare atoms in circular
Rydberg states by laser photoexcitation to approppriate `- and m`-mixed states which then evolve into circular states when the electric field strength is adiabatically reduced towards zero. The Hamiltonian associated with a single Rydberg electron in weak crossed electric F~ = (Fx, 0, 0) and magnetic
B~ = (0, 0, Bz) fields can be written in atomic units to first order as
2 p 1 Bz m` = + + Fx x, (2.16) H − 2 − r 2 | {z } | {z } c W H where is the Coulomb-interaction Hamiltonian and W the pertubation Hc due to the applied fields. The crossed-fields approach to the preparation of
37 circular states is based on the SO(4) symmetry of the Coulomb interaction,
c, with the generators Łij [87], where H
Łij = εijk Lk, · (2.17) Łi4 = Ai.
Here εijk is the fully antisymmetric tensor, ~L is the angular momentum vector and A~ is the Runge-Lenz vector. One suitable choice of eigenbasis for this Hamiltonian is the subgroup chain SO(4) SO(3) SO(2)m, ⊃ λ ⊃ which preserves the rotational invariance in real space. Their associated 2 eigenfunctions ~λ and λz, derived from the operator ~λ = (Ł14, Ł24, Ł12), are solutions to the Coulomb problem. The corresponding energy eigenvalues of the hydrogen atom in atomic units are
1 E(0) = (2.18) n −2n2
The perturbation due to the applied fields W can be diagonalized for each n atomic shell. After replacing the operator x = (3/2)n Ax [87], one obtains − ·
Wn = ωLLz + ωSAx q = ω2 + ω2~λ u~, (2.19) L S ·
Here ω = Bzm /2 and ω = ( 3/2)nFx are the Larmor and Stark frequencies L ` S − in atomic units, and u~ = (sin α, 0, cos α) can be interpreted as a rotation of
1 the operator ~λ in the xz-plane by the angle α = tan− (ωS/ωL).
From Eq. (2.18) and Eq. (2.19) it can be seen that the eigenfunctions of c, H (~λ)2, and W commute, so they can be chosen as the eigenfunctions of the total Hamiltonian, whose eigenenergies are q 1 2 2 En = + k ω + ω , (2.20) −2n2 L S 38 Figure 2.3: Adiabatically switching off the electric field, F~, leads to a rotation of the vector operator ~λ. This is equivalent to the evolution of the upper state’s parabolic ellipse, with angular momentum m = 0, to a
circular trajectory with m = Lz = n 1. − where k is the eigenvalue of the (λ u~)-operator. Circular Rydberg states · can therefore be prepared by starting out in the Stark limit ωS ωL, optically exciting the outermost Stark state with maximum eccentricity k = n 1 = λx = Ax, as illustrated in Fig. (2.3). This is possible because max − the projection of the angular momentum vector in the direction of the applied electric field is Lx = 0 = m. In this case the vector ~λ is directed along the x-axis (electric field axis), because α π , which is the angle ≈ 2 between the z-axis and ~λ. By slowly switching off the electric field, the system evolves adiabatically to the limit in which ω ω , i.e., α 0, and L S ≈ a circular Rydberg state λ = Lz = n 1 = m , while remaining in the − max upper sublevel, k = n 1, because the magnitude of the vector operator max − ~λ = n 1 is conserved since the upper sublevel is a coherent state of the | − i SO(3)λ subgroup. Using this crossed-fields method a circular Rydberg state results, as the electron evolves into the state with k , m . Even though this method | max maxi is described here for the case of the hydrogen atom, as long as the upper
39 state Lx = 0 can be selectively prepared by either laser photoexcitation from the ground state or by optical-microwave double resonance methods, it can be used to efficiently produce circular Rydberg states of any atomic species, such as helium [127]. It has also been applied in the preparation of Rydberg states in hydrogen [73], rubidium [16], cesium [59], and other atoms [82, 26, 5].
2.3.1 Helium Atoms in Circular Rydberg States
Helium atoms are well suited to hybrid quantum information processing with Rydberg atoms and transmission line resonators because of their low adsorption rates onto cryogenic chip surfaces which in turn minimizes the build-up of time-varying stray electric fields [109]. In these experiments it is desirable to operate the superconducting circuits at frequencies be- low 20 GHz. At such frequencies resonators with higher quality factors ∼ can be realized. To achieve this therefore requires atoms in circular Rydberg states with n 70. However, the preparation of circular Rydberg states ≥ with these high values of n is particular challenging due to the high sen- sitivity of the photoexcitation phase of the crossed-fields method to stray electric fields. For non-hydrogenic atoms such as helium the standard ap- proach to circular state preparation using the crossed-fields method is to initially excite the outermost positively shifted Stark state with m` = 0, with respect to the electron field quantization axis, at the first avoided crossing of this state with a low-` state at the Inglis-Teller limit. The advantage of this approach is that at this avoided crossing the target outer Stark state can possess a significant fraction of the character of the low-` state, mak-
40 Figure 2.4: The energy level diagrams for steps 2-4 in the preparation process of circular Rydberg states for n = 70 in Helium as described in the text. From [80]. ing laser photoexcitation more efficient. However, for high values of n, the avoided crossings of these sublevels can be so small that the transitions to the individual states are not resolvable during the laser photoexcitation. To overcome this problem, the usual crossed-fields method, introduced in the previous section, can be modified [80]. Regarding notation, the circular states for specific values of n are denoted
41 here by nc for values of m` = (n 1). To prepare the helium atoms | ±i ± − in high-n circular Rydberg states, e.g. the 70c circular state, a resonant | +i two-photon laser excitation via an intermediate level is first performed in a weak electric field to transfer atoms from the metastable triplet 2s state | i to the 73s Rydberg state. In the second step of the process illustrated in | i Fig. (2.4) the electric field strength is adiabatically increased to polarize the
73s0 atoms, where the prime indicates an `-mixed Stark state with in a | i nonzero electric field, such that they have a static electric dipole moment of similar magnitude and orientation as the target outer Stark state with n = 70, i.e., the 70c0 state. This is achieved in a field just beyond the first | +i avoided crossing of the 73s state in the Stark map of the triplet Rydberg | i states of He. Due to the large overlap of the wavefunctions of this state and the 70c0 state, the corresponding large electric dipole transition moment | +i allows for fast and efficient population transfer to the 70c0 state by apply- | +i ing a resonant microwave pulse. Once the 70c0 state has been prepared, | +i the fourth and last step of the circular state preparation process is to slowly reduce the electric field strength so that the atoms evolve adiabatically in the presence of the magnetic field into the final 70c circular state. The | +i transition between the 70c and 71c circular Rydberg states possesses | +i | +i an electric dipole transition moment of 3477ea0, allowing for single photon coupling strengths on the order of g/(2π) 5 MHz [80] to two-dimensional ≈ superconducting microwave resonators. Circular states, such as these, are particularly well suited for hybrid cavity QED because they possess long coherence times ( 100 ms), low sensi- ∼ tivity to stray electric fields and possess large electric dipole transition
42 moments and transition frequencies in the microwave regime. Therefore, in the theoretical studies of involving Rydberg-atom–resonator coupling in Chapters 5, 6 and 7 of this thesis, the interaction of these high-n circular Rydberg states will be considered.
43 Chapter 3
Electrostatic Dipole Interactions in Polarized Rydberg Gases
The preparation of circular Rydberg states using the crossed fields method described in Chapter 2 requires the initial laser photoexcitation of Rydberg states with strong linear Stark shifts, and hence large static electric dipole moments. These states are then adiabatically converted into circular states in a zero electric field. To achieve quantum-state–selective preparation of a large number of circular Rydberg atoms the applied fields must be care- fully controlled and interactions between the excited atoms in each phase of the preparation process must be minimized. When the circular prepara- tion process is complete the dominant type of interaction between pairs of
6 excited atoms is the van der Waals interaction that scales with R− , where R is the distance between the atoms. However, the large static electric dipole moments of the Rydberg-Stark states initially laser photoexcited in- teract through the stronger static electric dipole-dipole interactions, which
3 depend on R− . A wide range of studies, both theoretical and experimen-
44 tal, have been carried out on van der Waals and resonant dipole-dipole interactions between Rydberg atoms [35, 9, 34]. These interactions, and the associated excitation blockade mechanisms, are exploited in, e.g., the entanglement of pairs of atoms [91, 118], and in applications of Rydberg atoms in quantum simulation [61, 79]. However, there has been little com- parable study on the large static electric dipole interactions relevant to the circular state preparation processes discussed in Chapter 2. For this reason, and with the aim of elucidating the role of such interactions in Rydberg- Stark deceleration and trapping experiments, described in this chapter are the results of experiments and calculations of effects of static electric dipole interactions in dense gases of strongly polarized helium Rydberg atoms. The work described in this chapter was published in Zhelyazkova, Jirschik, and Hogan, Physical Review A 94.5: 053418 (2016) [128]. The text here follows closely the content of the published article, with adaptations to fit the structure of this thesis. The experiments described in Sections 3.1 and 3.2 were performed by Dr. Valentina Zhelyazkova while I carried out the calculations described in Section 3.3. Dr. Valentina Zhelyazkova and I both contributed to the preparation of the text of the published article.
The static electric dipole moments of Rydberg atoms and molecules can exceed 10000 D for states with values of n 50 and scale with n2. These ≥ dipole moments give rise to strong linear Stark shifts in external electric fields [36], and allow for control of the translational motion of a wide range of neutral atoms and molecules [54], composed of matter [123, 114, 52, 55] and antimatter [30], using inhomogeneous electric fields. When prepared
45 in states with these large dipole moments, gases of Rydberg atoms can also exhibit strong electrostatic dipole-dipole interactions [37, 125, 117]. These inter-particle interactions cause energy shifts of the Rydberg states, and, as demonstrated here, can modify the dielectric properties of the gases. The observation, and spectroscopic characterization, of the effects of these electrostatic interactions opens opportunities for using beams of Rydberg atoms or molecules as model systems with which to study many-body pro- cesses [23], including, e.g., resonant energy transfer [103], or surface ioniza- tion [50, 70, 40]. It is also of importance in the elucidation of effects of dipole interactions in Rydberg-Stark deceleration and trapping experiments [101], in experiments involving long-range Rydberg molecules [43, 10] possess- ing large static electric dipole moments [14], and in applications of Rydberg states as microscopic antennas for the detection of low-frequency electric fields [125, 78]. Microwave spectroscopy of the effects of van der Waals interactions [95, 86, 106], and resonant dipole-dipole interactions [2, 85] in cold Rydberg gases have been reported previously and provided important insights into these interacting few-particle systems. The electrostatic interactions that occur between Rydberg atoms in the outermost Stark states with the largest electric dipole moments, µ (3/2)n2ea , represent the extreme case of max ' 0 the dipole-dipole interaction for each value of n. The interaction potential, V , between a pair of atoms with electric dipole moments, µ = µ~ and dd 1 | 1| µ = µ~ , aligned with an electric field, F~, can be expressed as [37] 2 | 2|
µ1µ2 2 Vdd = 3 1 3 cos θ , (3.1) 4π0R − where R = R~ is the inter-atomic distance, θ is the angle between R~ and F~, | | 46 and 0 is the vacuum permittivity. For an isotropic, aligned ensemble of dipoles the average interaction energy is non zero, and leads to mean-field energy-level shifts within the ensemble. The microwave spectroscopic studies reported here represent a direct measurement of these mean-field shifts in gases of helium (He) excited to Rydberg states with n = 70 and electric dipole moments up to µ1 =
µ70 = 12 250 D. In the highest density regions of these gases, the dielectric properties differ from those in free space and give rise to changes in the microwave spectra which reflect the emergence of macroscopic electrical properties of the medium. This work is complimentary to recent studies of effects of electrostatic dipole interactions on particle motion in laser cooled samples of rubidium [108], and observations of optical bistability arising from mean-field shifts in gases of Rydberg atoms interacting via resonant dipole interactions [22].
3.1 Experiment
The experiments were performed in pulsed supersonic beams of metastable helium. The He atoms in these beams had a mean longitudinal speed of
1 1 2000 m s− with a spread of 50 m s− [126]. The atoms were prepared in ∼ ± 3 the metastable 1s2s S1 level in an electric discharge at the exit of a pulsed valve. This valve has operated at a repetition rate of 50 Hz [46]. After col- limation by a 2 mm diameter skimmer and the deflection of stray ions pro- duced in the discharge, the beam entered between a parallel pair of 70 mm 70 mm copper electrodes. The electrodes were separated by 8.4 mm × in the z dimension, as indicated in Fig. 3.1(a). At the mid-point between
47 (a) MCP V1 v He* z F 8.4 mm z 20 mm V2 x y (b) n = 70 n = 72 signal - e Time Ionization electric field
11 µs ~1 µs
Microwave pulse
∆tµ ~ 1 µs
Polarization electric field
Photoexcitation 200 ns 5 µs pulse
3 µs Time
Figure 3.1: (a) Schematic diagram of the photoexcitation and detection re- gion of the experimental apparatus. The position of laser photoexcitation and the approximate spatial extent of the Rydberg atom ensemble in the y dimension during microwave interrogation are indicated by the red circle and dashed rectangle, respectively. (b) Sequence of events in each experi- mental cycle, including laser photoexcitation, Rydberg state polarization, microwave interrogation, and state-selective electric field ionization (not to scale). From [128].
48 these electrodes the atomic beam was crossed at right-angles by two co- propagating, frequency stabilized, cw laser beams. These lasers were oper- ated at wavelengths of λUV = 388.975 nm and λIR = 785.946 nm to excite the metastable atoms to Rydberg states by the 1s2s 3S 1s3p 3P 1s70s 3S 1 → 2 → 1 two-photon excitation scheme. The laser beams were focussed to full- width-at-half-maximum (FWHM) beam waists of 100 µm. By applying ∼ a pulsed potential, V1, to the upper electrode in Fig. 3.1(a) to bring the atoms in the photoexcitation region into resonance with the lasers for 3 µs [see Fig. 3.1(b)], 6 mm-long ensembles of Rydberg atoms were generated in each cycle of the experiment. After photoexcitation, the electric field was switched again to polar-
3 ize the 1s70s S1 atoms [see Fig. 3.1(b)]. Interactions within the excited ensemble were then probed by microwave spectroscopy on the single pho- ton transition at 37 GHz between the states that evolve adiabatically to ∼ 3 3 the 1s70s S1 and 1s72s S1 levels in zero electric field. These states have quantum defects of δ δ 0.296 664 [32] and exhibit quadratic Stark 70s ' 72s ' energy shifts in weak electric fields. Because the laser photoexcitation process was separated in time from microwave interrogation, the Rydberg atom number density, nRy, could be adjusted by controlling the IR laser intensity, without affecting any other experimental parameters. After trav- eling 20 mm from the position of laser photoexcitation, the Rydberg atoms were detected by state-selective ramped-electric-field ionization beneath a 3 mm-diameter aperture in the upper electrode [see Fig. 3.1(a)]. This ionization step in the experiments was carried out by applying a slowly- rising ionization potential, V2, to the lower electrode. The electrons from
49 -630 n = 72 72p -640 72s (i) (ii) (iii)
-650 n = 71 71p -660 71s Energy / h (GHz)
-670 n = 70 70p 70s -680
0 0.2 0.4 0.6 0.8 1.0 1.2 Electric field (V/cm)
Figure 3.2: Stark structure of the triplet m` = 0 Rydberg states of He with values of n ranging from 70 to 72. Experiments were performed following laser photoexcitation of the 70s state in zero electric field and the subsequent polarization of the excited ensemble of atoms in fields
1 1 1 of (i) 18 mV cm− , (ii) 358 mV cm− , and (iii) 714 mV cm− as indicated. From [128].
50 the ionized Rydberg atoms were detected on a microchannel plate (MCP) detector located above this electrode. The rise time of the pulsed ionization electric field was selected to separate in time the electron signal from the n = 70 and n = 72 states.
3.2 Results
The mean-field energy-level shifts in the ensembles of polarized Rydberg gases were observed directly by microwave spectroscopy. To achieve this, reference measurements were first made in which the polarizing electric
1 1 field was switched after photoexcitation from 0 1 mV cm− to 18 mV cm− , ± 3 inducing electric dipole moments of µ70 = 380 D in the 1s70s S1 atoms. Here, and in the following, the dipole moments referred to represent the absolute value of the first derivative of the Stark energy shift with respect to the electric field. The n = 70 n = 72 transition (µ = 470 D), → 72 indicated by the dashed line labelled (i) in Fig. 3.2, was then driven by a 1 µs-long microwave pulse [see Fig. 3.1(b)]. The corresponding Fourier- transform-limited spectrum of the integrated electron signal associated with the n = 72 state is displayed in Fig. 3.3(a-i) (continuous curves). In order to identify and characterize the mean-field energy-level shifts in the ensembles of excited atoms measurements were performed at two Rydberg atom densities. The low (high) density data were recorded following the excitation of NRy (16 NRy) Rydberg atoms, and are displayed in blue (red) in the figure. For the small electric dipole moments induced in this case, no density-dependent energy-level shifts were observed. To enhance the effects of electrostatic dipole interactions within the
51 (a) (b) (iii) (iii) 3.0 3.0 µ = 14050 D Low High ) 72 )
µ70 = 12250 D density density 2.5 2.5
2.0 2.0 (ii) (ii) µ = 8650 D Low High 1.5 72 1.5 density density µ70 = 7250 D
1.0 1.0 (i) (i)
0.5 µ72 = 470 D Experiment 0.5 Experiment
µ70 = 380 D Calculation Calculation Integrated electron signal (arb. units Integrated electron signal (arb. units 0.0 0.0
-15 -10 -5 0 +5 +10 +15 -15 -10 -5 0 +5 +10 +15 Relative frequency (MHz) Relative frequency (MHz)
Figure 3.3: Experimentally recorded (continuous curves) and calculated (dashed curves) spectra of transitions between the Stark states that evolve
3 3 adiabatically to the 1s70s S1 and 1s72s S1 levels in zero electric field. The experiments were performed at low (blue curves) and high (red curves) Rydberg atom number densities, and spectra were recorded in polarizing
1 1 1 electric fields of (i) 18 mV cm− , (ii) 358 mV cm− , and (iii) 714 mV cm− , for which µ70 = 380 D, 7250 D, and 12250 D (µ72 = 470 D, 8650 D, and 14050 D), respectively. The calculated spectra are presented (a) without, and (b) with effects of the local polarization of the medium on the macroscopic dielectric properties accounted for. The relative microwave frequencies on the horizontal axes are displayed with respect to the transition frequencies [(i) 37.2413 GHz, (ii) 37.1068 GHz, and (iii) 36.8094 GHz] recorded at low Rydberg atom number density. From [128].
52 Rydberg gases, subsequent spectra were recorded with further polariza-
1 1 tion. Increasing the electric field to 358 mV cm− (714 mV cm− ) resulted in induced electric dipole moments of µ70 = 7250 D (µ70 = 12250 D), and
µ72 = 8650 D (µ72 = 14050 D). In the spectra recorded in these fields, and displayed in Fig. 3.3(a-ii) and (a-iii), respectively, a significant dependence on the Rydberg atom density was observed. Comparison of the spectra in Fig. 3.3(a-ii) and (a-iii) with those in Fig. 3.3(a-i) reveals four notable fea- tures. When the atoms are more strongly polarized the spectral profiles: (1) become significantly broader than those in Fig. 3.3(a-i), even at low density; (2) exhibit density-dependent frequency shifts; (3) become asym- metric at high density with a sharp cut-off in intensity at higher microwave transition frequencies; and (4) display signatures of spectral narrowing at high density, in particular in Fig. 3.3(a-iii).
3.3 Simulations
To aid in the interpretation of the spectra, Monte Carlo calculations were performed in which the electrostatic dipole-dipole interactions within the many-particle system were treated. In these calculations, ensembles of Rydberg atoms were generated with randomly assigned positions. These ensembles had Gaussian spatial distributions with a FWHM of 100 µm (50 µm) in the x (z) dimension, and flat top distributions with a length of 3 mm in the y dimension. These distributions represent the spatial intensity profile of the laser beams in the z dimension, the spatial distribution of atoms with Doppler shifts within the spectral width (∆ν 15 MHz) FWHM ' of the 1s2s 3S 1s3p 3P transition in the x dimension, and half of the 1 → 2 53 length of the ensemble of atoms excited in the y dimension. For each atom in the calculation the difference between the sum of the dipole-dipole interaction energies with all other atoms when it was (1) in the n = 70 state, i.e., µ1 = µ2 = µ70, and (2) in the n = 72 state, i.e., µ1 = µ72, and µ2 = µ70, was determined. To account for the sharp cut-off in the spectral profiles at high transition frequencies in the data recorded at high density in Fig. 3.3(a-ii) and (a-iii), it was necessary to impose a lower limit,
Rmin, on the nearest-neighbor spacing between pairs of Rydberg atoms in the calculations. This minimum inter-atomic spacing is a consequence of the expansion of the pulsed supersonic beams as they propagate from the valve to the photoexcitation region in the experiments, a distance of 210 mm, and reflects the collision-free environment characteristic of these beams [102, 71]. From a global fit to all of the experimental data in Fig. 3.3, carried out by calculating and subsequently minimizing the root mean- squared error between the experimental data and the models obtained for different values of Rmin, the most appropriate value of Rmin was found to be 11.5 µm.
3 We note that because the C6, van der Waals coefficient of the 1s70s S1 level in He is 9.9 GHz µm6, blockade at laser photoexcitation does not ∼ play a major role at the laser resolution in the experiments for atoms sep- arated by more then 3.5 µm and therefore does not dominate this value ∼ of R . Under these conditions, the distribution of n = 70 n = 72 tran- min → sition frequencies was determined for each set of electric dipole moments (i.e., for each electric field strength) for atoms located within 0.75 mm of ± the center of the ensemble in the y-dimension to avoid edge effects. The
54 resulting data were convoluted with a Gaussian spectral function with a FWHM of 1 MHz corresponding approximately to the Fourier transform of the microwave pulses. When the polarization of the individual atoms is increased they become more sensitive to low-frequency electrical noise [126]. This sensitivity led to the broadening of the resonances in the low-density regime in Fig. 3.3(a- ii) and (a-iii). The effect of this spectral broadening was introduced in the calculations as a perturbing electric field which caused a relative shift in the energy of the n = 70 and n = 72 Stark states, proportional to their electric dipole moments. The spectra calculated following the generation of low density samples containing NRy = 2 813 atoms and including the contributions from this electric field noise are displayed in Fig. 3.3(a) (blue dashed curves). Comparing the experimental data with the results of these calculations indicates that the broadening observed was commensurate
1 with white noise with a root-mean-square amplitude of Fnoise = 2 mV cm− . In the more polarized gases, increasing the Rydberg atom density is seen to shift the resonant microwave transition frequencies by +3 MHz ∼ and +5 MHz in Fig. 3.3(a-ii) and (a-iii), respectively, from those recorded ∼ at low density. These density-dependent changes indicate that atom-atom interactions dominate the spectral broadening caused by Fnoise. The mean- field shifts in transition frequency were found to agree with those seen in the results of the calculations upon increasing the number of excited atoms by a factor of 16, matching the experiments, to NRy = 45 000 [red dashed curves in Fig. 3.3(a)]. The positive shifts in transition frequency with increasing density are a consequence of the predominantly repulsive
55 electrostatic dipole interactions within the elongated ensemble of Rydberg atoms in the experiments, and the larger energy shift of the more polar, n = 72 state under these conditions. However, the spectral narrowing observed at high density in the ex- perimental data is not seen in the calculations in Fig. 3.3(a). To account for this it was necessary to consider the contribution of the local polarization of the Rydberg gases on their dielectric properties [84]. In a simple model of the dielectric Rydberg gas, the local polarization Ploc = nloc µ70 (nloc is the Rydberg atom number density obtained in the numerical calculation within a sphere of radius 25 µm surrounding each atom) can be consid- ered to shield each atom from the laboratory electrical noise, reducing it to F = max( 0, F S P / ), where S is a constant factor that accounts loc { noise − κ loc 0} κ for the geometry of the ensemble of atoms in the experiments. This shape factor is a fitting parameter associated with the proportion of local Rydberg atoms affected by the dielectric screening of the noise, which varies with
Rmin and accounts for the arbitrary choice of the radius of the sphere in which the local atom density is calculated. For Sκ = 0.1, there is excellent agreement between the results of the calculations and the experimental data as can be seen in Fig. 3.3(b). This indicates the emergence of macro- scopic electrical properties of the Rydberg gas when strongly polarized. In the experiments, free He+ ions, generated by photoionization of Ryd- berg atoms, would give rise to spectral broadening, or, if they increase the strength of the local electric fields within the ensembles of atoms, shifts of the spectral features toward lower transition frequencies (see Fig. 3.2). They are not expected to contribute to spectral narrowing such as that
56 4.0 (d) Experiment 3.5 Calculation 8 -3 nRy ≈ 3.3 × 10 cm 3.0 (c)
8 -3 2.5 nRy ≈ 2.8 × 10 cm
2.0 (b) n ≈ 2.1 × 108 cm-3 1.5 Ry
1.0 (a) n ≈ 7.5 × 107 cm-3 0.5 Ry Integrated electron signal (arb. units)
0.0
-15 -10 -5 0 +5 +10 +15 Relative frequency (MHz)
Figure 3.4: Experimentally recorded (continuous curves) and calculated (dashed curves) dependence of the n = 70 n = 72 spectra of the most → strongly polarized ensemble of atoms, for which µ70 = 12 250 D and
µ72 = 14050 D, on the Rydberg atom density. From [128].
57 observed in Fig. 3.3(a,ii-iii) and (b,ii-iii). For these reasons, and because we do not observe free ions or electrons in the process of ionization of the Rydberg atoms, we conclude that they do not contribute significantly to the experimental observations. Further comparisons of experimental and calculated spectra of the most polarized atoms [Fig. 3.3(b,iii)] over a range of Rydberg atom densities are presented in Fig. 3.4. From these data, it can be seen that the experimental spectra, recorded upon increasing NRy by factors of 5, 9.5 and 16 [Fig. 3.4(b), (c) and (d), respectively] from the initial low density case [Fig. 3.4(a)], agree well with the calculated spectra containing equivalent proportions of atoms, i.e., NRy = 2 813, 14 063, 26 719, and 45 000. From the results of the calculations the mean Rydberg atom number densities, nRy, within
7 3 the samples could be determined to range from n = 7.5 10 cm− to Ry × 8 3 3.3 10 cm− , as indicated. In these strongly polarized gases the labo- × ratory electric field noise, F , is completely screened when S P / noise κ loc 0 ≥ 1 8 3 2 mV cm− , i.e., when n 4.5 10 cm− , and the absolute energy-level loc ≥ × shift of each atom is 20 MHz. ∼
3.4 Conclusions
In conclusion, we have carried out spectroscopic studies of mean-field energy-level shifts in strongly polarized Rydberg gases by driving mi- crowave transitions between Rydberg-Stark states with similar electric dipole moments. The resulting spectra yield detailed information on the spatial distributions of the atoms. Spectral narrowing observed at high number density is attributed to changes in the local dielectric properties
58 within the medium from those in free space, and reflects the macroscopic electrical properties of the atomic samples that emerge under these con- ditions. From these results, it is seen that for atoms prepared in states with static electric dipole moments on the order of 10000 D dipole-dipole interactions result in frequency shifts in excess of 10 MHz for atom-atom separations of / 10 µm. To ensure that such frequency shifts remain below this value, and hence, are not detrimental to quantum-state–selective circu- lar state preparation the number densities of excited atoms in experiments
8 1 must be maintained below 10 cm− .
59 Chapter 4
Superconducting Transmission Line Resonator
In the development of hybrid approaches to quantum optics and quantum information processing superconducting circuits and qubits are uniquely scalable systems with high gate operation rates but short coherence times [115]. On the other hand, atoms in Rydberg states offer lower gate opera- tion rates for similar transition frequencies but, being in the gas phase, can possess significantly longer coherence times [97]. The key component re- quired for coupling these two qubits is a high quality microwave resonator. To maintain scalability, transmission line resonators offer particular advan- tages in such hybrid system setups. In this chapter the characteristic properties of superconducting transmis- sion line resonators are described. These properties are then employed in the theoretical studies in the following chapters.
60 Figure 4.1: Simple LC-circuit electrical oscillator: (a) In order to derive a Lagrangian where the single degree of freedom is the charge, q, the usual sign convention for the flow of the current, I, from q to q can be − used. (b) The sign convention in the text, where the positive and negative signs of the charge, Q, have been flipped with respect to the current, leads to the magnetic flow as the single degree of freedom in the Lagrangian describing the LC-circuit. From [41].
4.1 Quantization of a 1D Transmission Line Res-
onator
A one-dimensional transmission line resonator can be modelled as a series of harmonic LC-circuits. This electric circuit element consists of an inductor with inductance L and a capacitor with capacitance C. The resonant circuit is described by the potential energy, EC, stored in the capacitor
CV2 E = , (4.1) C 2
61 where V is the voltage across the capacitor, and the kinetic energy stored in the magnetic field, EL, which is induced by the current I, flowing through the inductor 1 E = LI2. (4.2) L 2 The inductor current I = Q˙ is the rate of change of capacitor charge with time. The magnetic flux Φ can be expressed in terms of the applied voltage using Faraday’s law of inductance,
Z t Φ(x, t) = dτ V(x, τ). (4.3) 0
Considering the definition of the inductance, i.e., L = Φ/I, the energy stored in the capacitor has the form of a kinetic energy and the energy stored in the inductor is analogous to the potential energy of a particle in one-dimensional space. Consequently, the Lagrangian for the LC-circuit can be written in terms of the magnetic flow as
1 2 1 2 LC = CΦ˙ Φ . (4.4) L 2 − 2L
The conjugate momentum to this single degree of freedom, Φ, is therefore
∂ L = CΦ˙ (4.5) ∂Φ˙ = Q, (4.6) and hence the capacitator charge Q. Thus, the Hamiltonian for the LC- circuit is derived as
HLC = QΦ˙ − L !2 (4.7) C dΦ 1 = + Φ2, 2 dt 2L
62 which is the electronic version of a single-mode harmonic oscillator. The analogous lumped circuit element model of a 1D transmission line is a series of linearly coupled LC-circuits. The Hamiltonian associated with the transmission line can therefore be written as a discretized sum of N inductively coupled LC-circuits
N !2 N X C dΦ X 1 2 HTL = + Φj Φj 1 . (4.8) 2 dt 2L − − j=1 j=2
This expression has the same form as a harmonic chain. In the continuum limit, a capacitance per unit length, c, and inductance per unit length, l, can be defined. For a transmission line of length d, the Lagrangian can then be written as [12] Z d c 2 1 2 TL = dx (∂tΦ) (∂xΦ) , (4.9) L 0 2 − 2l which leads to the Euler-Lagrange equation ! d2 1 ∂2 Φ = 0. (4.10) dt2 − lc ∂x2
This is the equivalent of a wave equation with velocity v = 1 . In order √lc that the transmission line acts as a resonator which stores an electromag- netic field in form of standing waves, series capacitances are implemented at each end of the transmission line to impose open-circuit boundary con- ditions such that
∂Φ ∂Φ = = 0. (4.11) ∂x x=0 ∂x x=d The boundary conditions are time-independent. Hence, the magnetic flux field can be factorized
X∞ Φ(x, t) = fn(t)φn(x) (4.12) n=0
63 where the amplitude fn(t) of each field mode n depends only on time and
φn(x) are the normalized spatial eigenfunctions, whose derivatives vanish at the boundaries. These normal modes are expressed as r 2 φ (x) = cos (k x) (4.13) n L n
nπ where kn = with n N. The properties of these normal modes, i.e., that L ∈ Z L dxφn(x)φm(x) = δnm, (4.14) and hence Z L 2 dx∂xφn(x)∂xφm(x) = knδnm, (4.15) permit their use as an orthonormal basis in which to diagonalize the La- grangian in Eq. (4.9),
X∞ c h 2 2i = (∂t fn) ωn f , (4.16) LTL 2 − n n=0 where the wavevector is defined as kn = ωn/v. This Lagrangian, and subsequently the Hamiltonian, constitute a sum of independent simple harmonic oscillator normal modes, where the momentum conjugate to the amplitude fn(t) is ∂ qn = L = c∂t fn. (4.17) ∂(∂t fn) This set of independent normal modes can be promoted to quantum op- h i erators by imposing the commutation relation qˆn, fˆm = i~δn m. Since the − , classical Hamiltonian, which follows from Eq. (4.16),
X∞ 1 c H = q + ω2 f 2 , (4.18) TL 2c n 2 n n n=0
64 has a form that resembles the free electromagnetic field, and its quantiza- tion can be performed in a similar fashion [116]. The quantum operators qˆn, fˆm are expressed in terms of creation and annihilation operators r ˆ ~ fn = aˆn† + aˆn , (4.19) 2ωnc and r ~ωnc qˆn = i aˆ† aˆn , (4.20) 2 n − lead to the quantization of the magnetic flux and charge density via Eq. (4.12) and (4.17). These operators obey the standard field commu- tation relation
h i X∞ qˆ(x), Φˆ (x0) = i~ φ(x)φ(x0) − n (4.21)
= i~δ(x x0). − − Ultimately, the quantum Hamiltonian for the transmission line resonator with finite length d is Z d 1 2 1 2 HTL = dx qˆ + (∂xφˆ ) , (4.22) 0 2c 2l and can therefore be simplified by expressing it in terms of the ladder operators, taking the familiar form,
X∞ 1 Hˆ = ~ω aˆ†aˆ + . (4.23) TL n 2 n=0
4.2 Dissipation from a 1D TransmissionLine Res-
onator
Consider a circuit quantum electrodynamic system composed of a super- conducting transmission-line resonator coupled to a co-planar waveguide
65 which acts as an input/output channel. We aim to derive an expression for the quality factor of the transmission line resonator using a full quan- tum treatment of the coupled resonator-waveguide system. The classical
Figure 4.2: The resonator-waveguide system and its lumped circuit ana- logue. The resonator and waveguide can be considered as a sum of harmonic LC-circuits, where coupling capactiances at each end impose boundary conditions.
Lagrangian of this system can be written as a sum of the resonator and waveguide Lagrangians, with the addition of the capacitive coupling be- tween them [64], i.e.,
= L + + , (4.24) LTot Res LWG LC where 2 n 1 1 X φν φν 1 1 ˙ 2 ˙ 2 − − ˙ 2 Res = Ccφ0 + crdzφν + Ccφn, (4.25) L 2 2 − lrdz 2 ν=1
2 N 1 1 X Φµ Φµ 1 1 ˙ 2 ˙ 2 − − ˙ 2 WG = CcΦ0 + cwdzΦµ + CcΦN, (4.26) L 2 2 − lwdz 2 µ=1
66 and
= C φ˙ n Φ˙ . (4.27) LC − c · · 0
Here Cc is the coupling capacitance at each end of the resonator denoted by the subscript r, cr,w and lr,w are the capacitance and inductance per unit length of the resonator, cr and lr, and waveguide, cw and lw, respectively, and φn (Φn) is the magnetic flux at each point n in the respective lumped circuit element sequences associated with the resonator (waveguide), as indicated in Fig. 4.2. These expressions can be significantly simplified following diagonalization. An important approximation in the derivation of the corresponding Hamiltonian is the assumption that
C c 1. (4.28) cr,wLr,w
This criterion is generally fulfilled since the coupling capacitances must be small to allow for an appropriate build-up and storage of an electromag- netic field inside the resonator. Writing the conjugate momenta ξn, Ξn of the variables qn, Qn leads to the Hamiltonian X X X X 1 h 2 2 2i 1 h 2 2 2i = q + ω ξ + Q + Ω Ξ + C ϕn(0)qn ψm(0)Qm. HTot 2 n n n 2 m m n c · n=0 m=0 n=0 m=0 (4.29) This expression describes two harmonic oscillators capacitively coupled at each end. Using the standard quantization procedure of a harmonic oscillator as shown in the previous section and by selectively exciting a single resonator mode, one obtains
X X ˆ = ~ωaˆ†aˆ + ~ Ωnbˆ†bˆn ~ gk bˆ† bˆk aˆ† aˆ , (4.30) H n − k − − n k=0
67 where the coupling strength gk can be expressed in terms of experimental parameters only, such that r r 1 1 1 1 p g = C ωΩ , k c c L q c L q k r r 1 + 2 Cc + C2Z2ω2 w w 1 + 2 Cc + C2Z2 Ω2 crLr c r cwLw c w k (4.31) where Zr,w is the impedence of the resonator and waveguide, respec- tively. It is known that the coupling strength should be proportional to the coupling capacitance and the impedance of the waveguide, since the influence of the transmission line on the resonator is generally due to its impedance [56]. In addition, the higher the energy of the mode ex- cited/supported by the resonator and waveguide, the higher the strength with which they couple to each other. Assuming they are made from the same material, and that all terms C2 are negligible, the coupling constant ∝ c in Eq.(4.31) then simplifies significantly to
2 Cc gk Z0π √k, (4.32) ≈ ωr with the the characteristic impedence of the transmission line Z0. This approximation of the coupling strength gk has the correct dimension of a frequency, i.e., Hz, and is in accordance with what we would expect for the simplified case described here. Following the standard quantum Langevin approach for a harmonic os- ciallator coupled to a collection of electromagnetic reservior oscillators, i.e., the waveguide taken to be in thermal equilibrium, under the RWA, one arrives at a decay rate [39] Z X dn γ = 2π g2δ (ω Ω ) 2π g2δ (ω Ω ) . (4.33) n n ∆n n n n − → −
68 The frequency modes of the waveguide are defined by a transcendental equation that follows from the wave-equation and boundary conditions of the Euler-Lagrange equations of motion. The solution of these equations can be obtained by a Taylor expansion up to first order, under the previous assumption Eq.(4.28), such that
π n Ωn = · . (4.34) 2 Cc + 1 L √c l cwLw w w w If the waveguide and resonator are considered to be composed of the same material, cr = cw = c and lr = lw = l, and the waveguide is a semi-infinite transmission line, L , for example, acting as a measurement channel, w → ∞ the resonator’s quality factor, Qext = ω/γ, can be written as